angular-cn/packages/compiler-cli/test/ngtsc/incremental_error_spec.ts

532 lines
16 KiB
TypeScript
Raw Normal View History

/**
* @license
* Copyright Google LLC All Rights Reserved.
*
* Use of this source code is governed by an MIT-style license that can be
* found in the LICENSE file at https://angular.io/license
*/
import {absoluteFrom as _} from '../../src/ngtsc/file_system';
import {runInEachFileSystem} from '../../src/ngtsc/file_system/testing';
import {loadStandardTestFiles} from '../../src/ngtsc/testing';
import {NgtscTestEnvironment} from './env';
const testFiles = loadStandardTestFiles();
runInEachFileSystem(() => {
describe('ngtsc incremental compilation with errors', () => {
let env!: NgtscTestEnvironment;
beforeEach(() => {
env = NgtscTestEnvironment.setup(testFiles);
env.enableMultipleCompilations();
env.tsconfig();
// This file is part of the program, but not referenced by anything else. It can be used by
// each test to verify that it isn't re-emitted after incremental builds.
env.write('unrelated.ts', `
export class Unrelated {}
`);
});
function expectToHaveWritten(files: string[]): void {
const set = env.getFilesWrittenSinceLastFlush();
perf(ivy): reuse prior analysis work during incremental builds (#34288) Previously, the compiler performed an incremental build by analyzing and resolving all classes in the program (even unchanged ones) and then using the dependency graph information to determine which .js files were stale and needed to be re-emitted. This algorithm produced "correct" rebuilds, but the cost of re-analyzing the entire program turned out to be higher than anticipated, especially for component-heavy compilations. To achieve performant rebuilds, it is necessary to reuse previous analysis results if possible. Doing this safely requires knowing when prior work is viable and when it is stale and needs to be re-done. The new algorithm implemented by this commit is such: 1) Each incremental build starts with knowledge of the last known good dependency graph and analysis results from the last successful build, plus of course information about the set of files changed. 2) The previous dependency graph's information is used to determine the set of source files which have "logically" changed. A source file is considered logically changed if it or any of its dependencies have physically changed (on disk) since the last successful compilation. Any logically unchanged dependencies have their dependency information copied over to the new dependency graph. 3) During the `TraitCompiler`'s loop to consider all source files in the program, if a source file is logically unchanged then its previous analyses are "adopted" (and their 'register' steps are run). If the file is logically changed, then it is re-analyzed as usual. 4) Then, incremental build proceeds as before, with the new dependency graph being used to determine the set of files which require re-emitting. This analysis reuse avoids template parsing operations in many circumstances and significantly reduces the time it takes ngtsc to rebuild a large application. Future work will increase performance even more, by tackling a variety of other opportunities to reuse or avoid work. PR Close #34288
2019-12-05 19:03:17 -05:00
const expectedSet = new Set<string>();
for (const file of files) {
perf(ivy): reuse prior analysis work during incremental builds (#34288) Previously, the compiler performed an incremental build by analyzing and resolving all classes in the program (even unchanged ones) and then using the dependency graph information to determine which .js files were stale and needed to be re-emitted. This algorithm produced "correct" rebuilds, but the cost of re-analyzing the entire program turned out to be higher than anticipated, especially for component-heavy compilations. To achieve performant rebuilds, it is necessary to reuse previous analysis results if possible. Doing this safely requires knowing when prior work is viable and when it is stale and needs to be re-done. The new algorithm implemented by this commit is such: 1) Each incremental build starts with knowledge of the last known good dependency graph and analysis results from the last successful build, plus of course information about the set of files changed. 2) The previous dependency graph's information is used to determine the set of source files which have "logically" changed. A source file is considered logically changed if it or any of its dependencies have physically changed (on disk) since the last successful compilation. Any logically unchanged dependencies have their dependency information copied over to the new dependency graph. 3) During the `TraitCompiler`'s loop to consider all source files in the program, if a source file is logically unchanged then its previous analyses are "adopted" (and their 'register' steps are run). If the file is logically changed, then it is re-analyzed as usual. 4) Then, incremental build proceeds as before, with the new dependency graph being used to determine the set of files which require re-emitting. This analysis reuse avoids template parsing operations in many circumstances and significantly reduces the time it takes ngtsc to rebuild a large application. Future work will increase performance even more, by tackling a variety of other opportunities to reuse or avoid work. PR Close #34288
2019-12-05 19:03:17 -05:00
expectedSet.add(file);
expectedSet.add(file.replace(/\.js$/, '.d.ts'));
}
perf(ivy): reuse prior analysis work during incremental builds (#34288) Previously, the compiler performed an incremental build by analyzing and resolving all classes in the program (even unchanged ones) and then using the dependency graph information to determine which .js files were stale and needed to be re-emitted. This algorithm produced "correct" rebuilds, but the cost of re-analyzing the entire program turned out to be higher than anticipated, especially for component-heavy compilations. To achieve performant rebuilds, it is necessary to reuse previous analysis results if possible. Doing this safely requires knowing when prior work is viable and when it is stale and needs to be re-done. The new algorithm implemented by this commit is such: 1) Each incremental build starts with knowledge of the last known good dependency graph and analysis results from the last successful build, plus of course information about the set of files changed. 2) The previous dependency graph's information is used to determine the set of source files which have "logically" changed. A source file is considered logically changed if it or any of its dependencies have physically changed (on disk) since the last successful compilation. Any logically unchanged dependencies have their dependency information copied over to the new dependency graph. 3) During the `TraitCompiler`'s loop to consider all source files in the program, if a source file is logically unchanged then its previous analyses are "adopted" (and their 'register' steps are run). If the file is logically changed, then it is re-analyzed as usual. 4) Then, incremental build proceeds as before, with the new dependency graph being used to determine the set of files which require re-emitting. This analysis reuse avoids template parsing operations in many circumstances and significantly reduces the time it takes ngtsc to rebuild a large application. Future work will increase performance even more, by tackling a variety of other opportunities to reuse or avoid work. PR Close #34288
2019-12-05 19:03:17 -05:00
expect(set).toEqual(expectedSet);
// Reset for the next compilation.
env.flushWrittenFileTracking();
}
it('should handle an error in an unrelated file', () => {
env.write('cmp.ts', `
import {Component} from '@angular/core';
@Component({selector: 'test-cmp', template: '...'})
export class TestCmp {}
`);
env.write('other.ts', `
export class Other {}
`);
// Start with a clean compilation.
env.driveMain();
env.flushWrittenFileTracking();
// Introduce the error.
env.write('other.ts', `
export class Other // missing braces
`);
const diags = env.driveDiagnostics();
expect(diags.length).toBe(1);
expect(diags[0].file!.fileName).toBe(_('/other.ts'));
expectToHaveWritten([]);
// Remove the error. /other.js should now be emitted again.
env.write('other.ts', `
export class Other {}
`);
env.driveMain();
expectToHaveWritten(['/other.js']);
});
it('should emit all files after an error on the initial build', () => {
// Intentionally start with a broken compilation.
env.write('cmp.ts', `
import {Component} from '@angular/core';
@Component({selector: 'test-cmp', template: '...'})
export class TestCmp {}
`);
env.write('other.ts', `
export class Other // missing braces
`);
const diags = env.driveDiagnostics();
expect(diags.length).toBe(1);
expect(diags[0].file!.fileName).toBe(_('/other.ts'));
expectToHaveWritten([]);
// Remove the error. All files should be emitted.
env.write('other.ts', `
export class Other {}
`);
env.driveMain();
expectToHaveWritten(['/cmp.js', '/other.js', '/unrelated.js']);
});
it('should emit files introduced at the same time as an unrelated error', () => {
env.write('other.ts', `
// Needed so that the initial program contains @angular/core's .d.ts file.
import '@angular/core';
export class Other {}
`);
// Clean compile.
env.driveMain();
env.flushWrittenFileTracking();
env.write('cmp.ts', `
import {Component} from '@angular/core';
@Component({selector: 'test-cmp', template: '...'})
export class TestCmp {}
`);
env.write('other.ts', `
export class Other // missing braces
`);
const diags = env.driveDiagnostics();
expect(diags.length).toBe(1);
expect(diags[0].file!.fileName).toBe(_('/other.ts'));
expectToHaveWritten([]);
// Remove the error. All files should be emitted.
env.write('other.ts', `
export class Other {}
`);
env.driveMain();
expectToHaveWritten(['/cmp.js', '/other.js']);
});
it('should emit dependent files even in the face of an error', () => {
env.write('cmp.ts', `
import {Component} from '@angular/core';
import {SELECTOR} from './selector';
@Component({selector: SELECTOR, template: '...'})
export class TestCmp {}
`);
env.write('selector.ts', `
export const SELECTOR = 'test-cmp';
`);
env.write('other.ts', `
// Needed so that the initial program contains @angular/core's .d.ts file.
import '@angular/core';
export class Other {}
`);
// Clean compile.
env.driveMain();
env.flushWrittenFileTracking();
env.write('cmp.ts', `
import {Component} from '@angular/core';
@Component({selector: 'test-cmp', template: '...'})
export class TestCmp {}
`);
env.write('other.ts', `
export class Other // missing braces
`);
const diags = env.driveDiagnostics();
expect(diags.length).toBe(1);
expect(diags[0].file!.fileName).toBe(_('/other.ts'));
expectToHaveWritten([]);
// Remove the error. All files should be emitted.
env.write('other.ts', `
export class Other {}
`);
env.driveMain();
expectToHaveWritten(['/cmp.js', '/other.js']);
});
it('should recover from an error in a component\'s metadata', () => {
env.write('test.ts', `
import {Component} from '@angular/core';
@Component({selector: 'test-cmp', template: '...'})
export class TestCmp {}
`);
// Start with a clean compilation.
env.driveMain();
env.flushWrittenFileTracking();
// Introduce the error.
env.write('test.ts', `
import {Component} from '@angular/core';
@Component({selector: 'test-cmp', template: ...}) // invalid template
export class TestCmp {}
`);
const diags = env.driveDiagnostics();
expect(diags.length).toBeGreaterThan(0);
expectToHaveWritten([]);
// Clear the error and verify that the compiler now emits test.js again.
env.write('test.ts', `
import {Component} from '@angular/core';
@Component({selector: 'test-cmp', template: '...'})
export class TestCmp {}
`);
env.driveMain();
expectToHaveWritten(['/test.js']);
});
it('should recover from an error in a component that is part of a module', () => {
// In this test, there are two components, TestCmp and TargetCmp, that are part of the same
// NgModule. TestCmp is broken in an incremental build and then fixed, and the test verifies
// that TargetCmp is re-emitted.
env.write('test.ts', `
import {Component} from '@angular/core';
@Component({selector: 'test-cmp', template: '...'})
export class TestCmp {}
`);
env.write('target.ts', `
import {Component} from '@angular/core';
@Component({selector: 'target-cmp', template: '<test-cmp></test-cmp>'})
export class TargetCmp {}
`);
env.write('module.ts', `
import {NgModule} from '@angular/core';
import {TargetCmp} from './target';
import {TestCmp} from './test';
@NgModule({
declarations: [TestCmp, TargetCmp],
})
export class Module {}
`);
// Start with a clean compilation.
env.driveMain();
env.flushWrittenFileTracking();
// Introduce the syntactic error.
env.write('test.ts', `
import {Component} from '@angular/core';
@Component({selector: ..., template: '...'}) // ... is not valid syntax
export class TestCmp {}
`);
const diags = env.driveDiagnostics();
expect(diags.length).toBeGreaterThan(0);
expectToHaveWritten([]);
// Clear the error and trigger the rebuild.
env.write('test.ts', `
import {Component} from '@angular/core';
@Component({selector: 'test-cmp', template: '...'})
export class TestCmp {}
`);
env.driveMain();
expectToHaveWritten([
// The file which had the error should have been emitted, of course.
'/test.js',
// Because TestCmp belongs to a module, the module's file should also have been
// re-emitted.
'/module.js',
// Because TargetCmp also belongs to the same module, it should be re-emitted since
// TestCmp's elector may have changed.
'/target.js',
]);
});
it('should recover from an error even across multiple NgModules', () => {
// This test is a variation on the above. Two components (CmpA and CmpB) exist in an NgModule,
// which indirectly imports a LibModule (via another NgModule in the middle). The test is
// designed to verify that CmpA and CmpB are re-emitted if somewhere upstream in the NgModule
// graph, an error is fixed. To check this, LibModule is broken and then fixed in incremental
// build steps.
env.write('a.ts', `
import {Component} from '@angular/core';
@Component({selector: 'test-cmp', template: '...'})
export class CmpA {}
`);
env.write('b.ts', `
import {Component} from '@angular/core';
@Component({selector: 'target-cmp', template: '...'})
export class CmpB {}
`);
env.write('module.ts', `
import {NgModule} from '@angular/core';
import {LibModule} from './lib';
import {CmpA} from './a';
import {CmpB} from './b';
@NgModule({
imports: [LibModule],
exports: [LibModule],
})
export class IndirectModule {}
@NgModule({
declarations: [CmpA, CmpB],
imports: [IndirectModule],
})
export class Module {}
`);
env.write('lib.ts', `
import {Component, NgModule} from '@angular/core';
@Component({
selector: 'lib-cmp',
template: '...',
})
export class LibCmp {}
@NgModule({
declarations: [LibCmp],
exports: [LibCmp],
})
export class LibModule {}
`);
// Start with a clean compilation.
env.driveMain();
env.flushWrittenFileTracking();
// Introduce the error in LibModule
env.write('lib.ts', `
import {Component, NgModule} from '@angular/core';
@Component({
selector: 'lib-cmp',
template: '...',
})
export class LibCmp {}
@NgModule({
declarations: [LibCmp],
exports: [LibCmp],
})
export class LibModule // missing braces
`);
// env.driveMain();
const diags = env.driveDiagnostics();
expect(diags.length).toBeGreaterThan(0);
expectToHaveWritten([]);
// Clear the error and recompile.
env.write('lib.ts', `
import {Component, NgModule} from '@angular/core';
@Component({
selector: 'lib-cmp',
template: '...',
})
export class LibCmp {}
@NgModule({
declarations: [LibCmp],
exports: [LibCmp],
})
export class LibModule {}
`);
env.driveMain();
expectToHaveWritten([
// Both CmpA and CmpB should be re-emitted.
'/a.js',
'/b.js',
// So should the module itself.
'/module.js',
// And of course, the file with the error.
'/lib.js',
]);
});
describe('chained errors', () => {
it('should remember a change to a TS file across broken builds', () => {
// Two components, an NgModule, and a random file.
writeTwoComponentSystem(env);
writeRandomFile(env, 'other.ts');
// Start with a clean build.
env.driveMain();
env.flushWrittenFileTracking();
// Update ACmp to have a different selector, isn't matched in BCmp's template.
env.write('a.ts', `
import {Component} from '@angular/core';
@Component({selector: 'not-a-cmp', template: '...'})
export class ACmp {}
`);
// Update the file to have an error, simultaneously.
writeRandomFile(env, 'other.ts', {error: true});
// This build should fail.
const diags = env.driveDiagnostics();
expect(diags.length).not.toBe(0);
expectToHaveWritten([]);
// Fix the error.
writeRandomFile(env, 'other.ts');
// Rebuild.
env.driveMain();
// If the compiler behaves correctly, it should remember that 'a.ts' was updated before, and
// should regenerate b.ts.
expectToHaveWritten([
// Because they directly changed
'/other.js',
'/a.js',
// Bcause they depend on a.ts
'/b.js',
'/module.js',
]);
});
it('should remember a change to a template file across broken builds', () => {
// This is basically the same test as above, except a.html is changed instead of a.ts.
// Two components, an NgModule, and a random file.
writeTwoComponentSystem(env);
writeRandomFile(env, 'other.ts');
// Start with a clean build.
env.driveMain();
env.flushWrittenFileTracking();
// Invalidate ACmp's template.
env.write('a.html', 'Changed template');
// Update the file to have an error, simultaneously.
writeRandomFile(env, 'other.ts', {error: true});
// This build should fail.
const diags = env.driveDiagnostics();
expect(diags.length).not.toBe(0);
expectToHaveWritten([]);
// Fix the error.
writeRandomFile(env, 'other.ts');
// Rebuild.
env.flushWrittenFileTracking();
env.driveMain();
// If the compiler behaves correctly, it should remember that 'a.html' was updated before,
// and should regenerate a.js. Because the compiler knows a.html is a _resource_ dependency
// of a.ts, it should only regenerate a.js and not its module and dependent components (as
// it would if a.ts were itself changed like in the test above).
expectToHaveWritten([
// Because it directly changed.
'/other.js',
// Because a.html changed
perf(ivy): reuse prior analysis work during incremental builds (#34288) Previously, the compiler performed an incremental build by analyzing and resolving all classes in the program (even unchanged ones) and then using the dependency graph information to determine which .js files were stale and needed to be re-emitted. This algorithm produced "correct" rebuilds, but the cost of re-analyzing the entire program turned out to be higher than anticipated, especially for component-heavy compilations. To achieve performant rebuilds, it is necessary to reuse previous analysis results if possible. Doing this safely requires knowing when prior work is viable and when it is stale and needs to be re-done. The new algorithm implemented by this commit is such: 1) Each incremental build starts with knowledge of the last known good dependency graph and analysis results from the last successful build, plus of course information about the set of files changed. 2) The previous dependency graph's information is used to determine the set of source files which have "logically" changed. A source file is considered logically changed if it or any of its dependencies have physically changed (on disk) since the last successful compilation. Any logically unchanged dependencies have their dependency information copied over to the new dependency graph. 3) During the `TraitCompiler`'s loop to consider all source files in the program, if a source file is logically unchanged then its previous analyses are "adopted" (and their 'register' steps are run). If the file is logically changed, then it is re-analyzed as usual. 4) Then, incremental build proceeds as before, with the new dependency graph being used to determine the set of files which require re-emitting. This analysis reuse avoids template parsing operations in many circumstances and significantly reduces the time it takes ngtsc to rebuild a large application. Future work will increase performance even more, by tackling a variety of other opportunities to reuse or avoid work. PR Close #34288
2019-12-05 19:03:17 -05:00
'/a.js', '/module.js',
// b.js and module.js should not be re-emitted, because specifically when tracking
// resource dependencies, the compiler knows that a change to a resource file only affects
// the direct emit of dependent file.
]);
});
});
});
});
/**
* Two components, ACmp and BCmp, where BCmp depends on ACmp.
*
* ACmp has its template in a separate file.
*/
export function writeTwoComponentSystem(env: NgtscTestEnvironment): void {
env.write('a.html', 'This is the template for CmpA');
env.write('a.ts', `
import {Component} from '@angular/core';
@Component({selector: 'a-cmp', templateUrl: './a.html'})
export class ACmp {}
`);
env.write('b.ts', `
import {Component} from '@angular/core';
@Component({selector: 'b-cmp', template: '<a-cmp></a-cmp>'})
export class BCmp {}
`);
env.write('module.ts', `
import {NgModule} from '@angular/core';
import {ACmp} from './a';
import {BCmp} from './b';
@NgModule({
declarations: [ACmp, BCmp],
})
export class Module {}
`);
}
export function writeRandomFile(
env: NgtscTestEnvironment, name: string, options: {error?: true} = {}): void {
env.write(name, `
// If options.error is set, this class has missing braces.
export class Other ${options.error !== true ? '{}' : ''}
`);
}